Cocaine Use Prediction With Tensor-Based Machine Learning on Multimodal MRI Connectome Data.

This letter considers the use of machine learning algorithms for predicting cocaine use based on magnetic resonance imaging (MRI) connectomic data. The study used functional MRI (fMRI) and diffusion MRI (dMRI) data collected from 275 individuals, which was then parcellated into 246 regions of interest (ROIs) using the Brainnetome atlas. After data preprocessing, the data sets were transformed into tensor form. We developed a tensor-based unsupervised machine learning algorithm to reduce the size of the data tensor from 275 (individuals) × 2 (fMRI and dMRI) × 246 (ROIs) × 246 (ROIs) to 275 (individuals) × 2 (fMRI and dMRI) × 6 (clusters) × 6 (clusters). This was achieved by applying the high-order Lloyd algorithm to group the ROI data into six clusters. Features were extracted from the reduced tensor and combined with demographic features (age, gender, race, and HIV status). The resulting data set was used to train a Catboost model using subsampling and nested cross-validation techniques, which achieved a prediction accuracy of 0.857 for identifying cocaine users. The model was also compared with other models, and the feature importance of the model was presented. Overall, this study highlights the potential for using tensor-based machine learning algorithms to predict cocaine use based on MRI connectomic data and presents a promising approach for identifying individuals at risk of substance abuse.

Longitudinal changes in neurocognitive performance related to drug use intensity in a sample of persons with and without HIV who use illicit stimulants.

BACKGROUND: Illicit stimulant use remains a public health concern that has been associated with multiple adverse outcomes, including cognitive deficits. The effects of stimulant use on cognition may be particularly deleterious in persons with HIV. Stimulant use intensity may be an important factor in the magnitude of observed deficits over time. METHODS: We completed neurocognitive testing in a sample of people who use stimulants with (n = 84) and without HIV (n = 123) at baseline and up to 4 follow-up time points over approximately 1 year. Participants reported on substance use at each visit, including frequency of use and stimulant dependence. Mixed effects models examined the relationship between stimulant-related factors and neurocognitive function over time. RESULTS: Participants were mostly male (57%), African American (86%), and 47.41 years old on average. All participants actively used stimulants at enrollment and use remained prevalent throughout the follow-up period, with an average of ≥24 days of use in the past 90 days at all time points. Retention was excellent, with 86% completing all 4 follow-up assessments. Mixed effects models showed that stimulant dependence was associated with lower neurocognitive performance independent of HIV status (p = 0.002), whereas frequency of use had a greater negative impact on performance in participants with HIV compared to those without HIV (p = 0.045). CONCLUSIONS: Our key finding is that stimulant-related factors are associated with neurocognitive performance over time, but in complex ways. These findings have important implications for harm reduction approaches, particularly those that target cognitive function.